基于GA的BP网络算法优化及应用

Optimization and Application of BP Network Algorithms Based on GA

BP(Back Propagation,BP)神经网络在日常生活中的应用非常广泛。BP神经网络简单实用,执行效率较高,但同样也存在收敛速度慢、会陷入局部极小值、容易出现“过拟合”的不足。而遗传算法(Genetic Algorithm,GA)具有极强的全域搜索能力,能快速地找到BP神经网络的最优解,即最优权值和阈值。用遗传算法来进行前期的搜索查找,能有效地克服BP算法的缺点。因此,将遗传算法GA与BP网络算法完美结合,具有很强的现实意义。为此,本文利用遗传算法改进的BP网络对矿井最优通风量进行合理预测。基于遗传算法优化的BP神经网络可以基于BP网络的特性高效地对数据进行处理分析,且具有适应性强、网络稳定度高的优势。

BP(back propagation) neural network is widely used in daily life. BP neural network is simple and practical, and its execution efficiency is high. However, it also has the disadvantages of slow convergence, falling into local minimum, and prone to “overfitting”.Genetic algorithm has strong global search ability and can quickly find the optimal solution of BP neural network, i.e. the optimal weight and threshold. Genetic algorithm can effectively overcome the shortcomings of BP algorithm. Therefore, the perfect combination of genetic algorithm GA and BP network algorithm has strong practical significance. Therefore, this paper uses BP network improved by genetic algorithm to reasonably predict the optimal ventilation volume of the mine. BP neural network optimized by genetic algorithm can process and analyze data efficiently based on the characteristics of BP network, and has the advantages of strong adaptability and high network stability.